echo-state-network

This paper benchmarks the connectomes of Caenorhabditis elegans using the reservoir computing framework, revealing that biological neural networks may not outperform randomized models in computational tasks.

This paper investigates the use of reservoir computing, specifically echo state networks, to predict nonlinear oscillations in a jumping quarter-car model, demonstrating feasibility for data-driven prediction of chaotic vehicle dynamics in smart driving applications.

Introduces a Lindblad-inspired multi-timescale reservoir architecture that separates rotation and dissipation for independent control of mixing, memory, and stability, achieving competitive results on benchmarks like NARMA-20 and Lorenz-63.